MSLR-Bench

With the rapid development of large language models (LLMs) in legal applications, systematically evaluating their reasoning ability in judgment prediction has become increasingly urgent. Currently, publicly available legal evaluation benchmarks lack a unified evaluation framework and do not adequately support these two tasks. To fill this gap, we propose MSLR, addressing a critical gap in structured reasoning evaluation in the field of Chinese legal natural language processing, and providing a solid foundation for the evaluation and optimization of legal vertical domain large model systems. For more details, please refer to our paper.